Pancreatic carcinoma usually presents at a clinical stage too advanced for surgical resection. Even with resection, the disease is usually eventually fatal. The precursor lesion for infiltrating adenocarcinoma of the pancreas is the pancreatic intraepithelial neoplasm, commonly termed a "hyperplasia". It is probably very common, being present in up to 60% of the older population. Only a subset are histologically advanced, and a small minority of persons harboring them will develop a cancer. Therefore, there is a need to reinterpret pancreatic carcinogenesis as a progression of intraepithelial neoplasia. Specifically, we need to know which markers define a "high risk" intraepithelial lesion that is likely to become invasive. The finding of K-ras mutations in a large number of these lesions has suggested that additional genetic markers need to be applied. Important questions remain unanswered, such as the multifocality of duct lesions, their physical extension along pancreatic ducts, which genes are altered in various stages of the lesions, whether the exact genetic changes in a cancer can be traced to a particular duct of origin, and whether markers of high risk lesions might be adapted for screening strategies. Preliminary studies along two lines have converged to allow the study of intraepithelial lesions. First, a number of genetic alterations have been defined in infiltrating pancreatic carcinomas, providing candidate genes for study of the precursors. Second, techniques for the study of small cell populations, including micromanipulator dissection to obtain individual cells, have been developed. Using these, we demonstrated an association of K-ras mutations with histologically advanced intraepithelial proliferations. Furthermore, we observed duct lesions having K-ras and p16 alterations identical to those of the resultant carcinomas - molecular confirmation for this tumor-progression model and suggestive that p16 mutations mark some high risk duct lesions. The specific aims of this project are to study the timing of specific gene mutations, characterize the extent and number of duct lesions among various population subgroups, identify and study the particular ducts which give rise to infiltrating carcinomas. The overall goal of this effort is therefore to define biologically "high risk" lesions, and explore the applicability to gene detection as a clinical screening tool.